arXiv:astro-ph/0208289v1 15 Aug 2002
An X-ray investigation of the NGC 346 field in the SMC (1) : the
LBV HD 5980 and the NGC 346 cluster
Y. Naz´e
1,8, J.M. Hartwell
2, I.R. Stevens
2, M. F. Corcoran
3, Y.-H. Chu
4, G. Koenigsberger
5,
A.F.J. Moffat
6, V.S. Niemela
7ABSTRACT
We present results from a Chandra observation of the NGC 346 cluster. This cluster contains numerous massive stars and is responsible for the ionization of N66, the most luminous H ii region and the largest star formation region in the SMC. In this first paper, we will focus on the characteristics of the main objects of the field. The NGC 346 cluster itself shows only relatively faint X-ray emission (with Lunabs
X ∼1.5 × 10
34 erg s−1), tightly correlated with the core of the
cluster. In the field also lies HD 5980, a LBV star in a binary (or possibly a triple system) that is detected for the first time at X-ray energies. The star is X-ray bright, with an unabsorbed luminosity of Lunabs
X ∼1.7 × 10
34 erg s−1, but needs to be monitored further to investigate its
X-ray variability over a complete 19 d orbital cycle. The high X-ray luminosity may be associated either with colliding winds in the binary system or with the 1994 eruption. HD 5980 is surrounded by a region of diffuse X-ray emission, which is a supernova remnant. While it may be only a chance alignment with HD 5980, such a spatial coincidence may indicate that the remnant is indeed related to this peculiar massive star.
Subject headings: (galaxies:) Magellanic Clouds–X-rays: individual (NGC 346, HD 5980)
1Institut d’Astrophysique et de G´eophysique, Universit´e de Li`ege, All´ee du 6 Aoˆut 17, Bat. B5c, B 4000 - Li`ege (Belgium); naze@astro.ulg.ac.be
2
School of Physics & Astronomy, University of Birmingham, Edgbaston, Birmingham B15 2TT (UK); jmh@star.sr.bham.ac.uk, irs@star.sr.bham.ac.uk
3
Universities Space Research Association, High En-ergy Astrophysics Science Archive Research Center, God-dard Space Flight Center, Greenbelt, MD 20771; corco-ran@barnegat.gsfc.nasa.gov
4
Astronomy Department, University of Illinois, 1002 W. Green Street, Urbana, IL 61801; chu@astro.uiuc.edu
5Instituto de Astronom´ıa, Universidad Nacional Aut´onoma de Mexico, Apdo. Postal 70-264, 04510, Mexico D.F. (Mexico); gloria@astroscu.unam.mx
6
D´epartement de physique, Universit´e de Montreal, C.P. 6128, Succ. Centre-Ville, Montreal, QC, H3C 3J7 (Canada); moffat@astro.umontreal.ca
7Facultad de Ciencias Astron´omicas y Geof´ısicas, Universidad Nacional de la Plata, Paseo del Bosque S/N, B19000FWA La Plata (Argentina); virpi@fcaglp.fcaglp.unlp.edu.ar
8
Research Fellow F.N.R.S.
1. Introduction
The Small Magellanic Cloud (SMC) is an irreg-ular galaxy at a distance of 59 kpc (Mathewson, Ford, & Visvanathan 1986) that forms a pair with the Large Magellanic Cloud (LMC). Both are satellites of our own Galaxy. The interstellar extinction towards the Magellanic Clouds is low, allowing studies of the X-ray sources to be under-taken with clarity.
Previous X-ray observations of the SMC have been made with the Einstein Observatory, ASCA and ROSAT. These observations included surveys of the point source population (Kahabka et al. 1999; Schmidtke et al. 1999; Haberl et al. 2000; Sasaki, Haberl, & Pietsch 2000; Yokogawa et al. 2000) and in particular studies of the properties of the X-ray binary population (Haberl & Sasaki 2000). The recent launch of Chandra, however, provides an opportunity to study the SMC with a far greater sensitivity and spatial resolution than
ever before.
We have obtained a Chandra observation of the giant H ii region N66 (Henize 1956), the largest star formation region in the SMC, in order to study the young cluster NGC 346 and its interaction with the surrounding interstellar medium. This cluster contains numerous massive stars (Massey, Parker & Garmany 1989), of spec-tral types as early as O2 (Walborn et al. 2002). The large number of massive stars is not the only feature of interest in this field. On the outskirts of NGC 346 lies the remarkable star HD 5980, which underwent a Luminous Blue Variable (LBV)-type eruption in 1994. This massive binary (or triple?) system has been monitored and analysed for its varying spectral and photometric properties in visible and UV wavelengths since the early 80’s (Koenigsberger et al. 2000; Sterken & Breysacher 1997, and references therein). Previous X-ray ob-servations of NGC 346 (Inoue, Koyama, & Tanaka 1983; Haberl et al. 2000) detected a bright ex-tended source around HD 5980 which has been attributed to a supernova remnant (SNR); how-ever, the crude instrumental resolution prohibited unambiguous detections of a point source associ-ated with HD 5980. In addition to HD 5980, the young massive stars of NGC 346 and their inter-acting winds are likely to produce X-ray emission, but have not been detected previously. The sen-sitive, high-resolution Chandra observation of the NGC 346 field thus provides an excellent oppor-tunity to study a variety of phenomena involving emission at X-ray energies.
In this paper, we will describe in § 2 the ob-servations used in this study, then discuss the available data on NGC 346, HD 5980, and the ex-tended emission in § 3, 4, and 5, respectively. Finally, conclusions are given in § 6. The second part of this work (Naz´e et al., paper II) will de-scribe the properties of the other sources detected in the field.
2. The Observations and Data Analysis 2.1. X-ray Observations
NGC 346 was observed with Chandra for the XMEGA9
consortium on 2001 May 15–16 for 100 ks (98.671 ks effective, ObsID = 1881, JD∼2 452 045.2d). The data were recorded by the ACIS-I0, I1, I2, I3, S2 and S3 CCD chips maintained at a temperature of −120◦C. The data
were taken with a frame time of 3.241s in the faint mode. The exposure was centered on the cluster, with HD 5980 lying 1.9′ to the north-east of the
aimpoint on ACIS-I3. Our faintest sources have fluxes of about 2 × 1032 erg s−1 (see paper II),
assuming a distance of 59 kpc for the SMC. We excluded bad pixels from the analysis, us-ing the customary bad-pixel file provided by the Chandra X-ray Center (CXC) for this particular observation. We have searched the data for back-ground flares, which are known to affect Chandra data, by examining the lightcurve of the total count rate, but no flares were found. Event Pulse Invariant (PI) values and photon energies were determined using the FITS Embedded File (FEF) acisD2000-01-29fef piN0001.fits.
For long exposures, removing the afterglow events can adversely affect the science analysis (underestimation of the fluxes, alteration of the spectra and so on)10. We thus computed a new
level 2 events file by filtering the level 1 file and keeping the events with ASCA grades of 0, 2, 3, 4 and 6, but without applying a status=0 filter. Throughout this paper, we will use this new file for all scientific analysis.
Further analysis was performed using the CIAO v2.1.2 software provided by the CXC and also with the FTOOLS tasks. The spectra were anal-ysed and fitted within XSPEC v11.0.1 (Arnaud 1996). In Fig. 1, we show a 3 color image of the Chandra data of the NGC 346 cluster, HD 5980 and its surroundings. We will discuss the features of this image further in later sections.
9
http://lheawww.gsfc.nasa.gov/users/corcoran/xmega/xmega.html 10
We have investigated problems due to Charge Transfer Inefficiency (CTI). To remove these CTI effects, we used the algorithm of Townsley et al. (2000). Checking several sources on different chips and/or at different positions on each CCD, we con-cluded that the impact of the CTI was small and that it did not change significantly the spectral parameters found by using non-corrected data. When a difference was apparently found, we dis-covered that the best fits found on corrected and non-corrected data gave similar χ2
. As the work of Townsley et al. (2000)is not yet an official CXC product11, we chose to present here only the
results of the non-corrected data.
2.2. HST WFPC2 Images
Two HST WFPC2 observations of the N66 / NGC 346 region in Hα are available from the STScI archives. They were taken on 1998 Sept. 27 and 2000 August 7 for the programs 6540 and 8196, respectively. HD 5980 is centered on the PC in the observations from program 6540 (see Fig. 2), while program 8196 shows an area situated fur-ther south. The observations were made through the F 656N filter for 2070s (Prog. # 6540) and 2400s (Prog. # 8196). This filter, centered at 6563.7 ˚A with a FWHM of 21.4˚A, includes the Hα line and the neighboring [N ii] lines.
The calibrated WFPC2 images were produced by the standard HST pipeline. We processed them further with IRAF and STSDAS routines. The images of each program were combined to remove cosmic rays and to produce a total-exposure map. No obvious correlation between X-rays features and the Hα emission is seen in the HST data. Fig. 2 will be discussed more extensively in § 5.
3. The NGC 346 Star Cluster
The NGC 346 cluster has a large number of massive stars, containing almost 50% of all early-type O stars in the SMC (Massey, Parker & Gar-many 1989). It is responsible for the ionization of N66, the most luminous H ii region in the SMC (Ye, Turtle, & Kennicutt 1991). However, the
11
http://www.astro.psu.edu/users/townsley/cti/
NGC 346 cluster has not been detected in X-rays until this Chandra observation. Even though NGC 346 lies near the joint of the four ACIS-I CCD chips and part of the cluster is lost in the gaps between CCDs, the X-ray emission from the core of NGC 346 is unambiguously detected in the corner of the ACIS-I3 chip. The X-ray emission appears extended with multiple peaks correspond-ing to the bright blue stars MPG 396, 417, 435, 470, and 476 (Fig. 3). As only ∼300 counts are de-tected and the X-ray peaks are not well-resolved, we will analyze only the overall emission from the cluster.
The total count rate of this emission in the 0.3 − 10.0 keV band is 3.04 ± 0.36 × 10−3cnts s−1.
The spectrum of the NGC 346 cluster was ex-tracted from a square aperture of ∼45′′×45′′with
a carefully chosen background region, as close as possible to the cluster. As we regard the clus-ter X-ray emission to be extended, we have used the calcrmf and calcarf tools to generate the ap-propriate response files. The best fit (χ2
ν=0.90,
Ndof=50, Z = 0.1Z⊙) to this spectrum was an
absorbed mekal model (Kaastra 1992) with the following properties (see Fig. 4): an absorption column density N (H) of 0.420.83
0.17×1022cm−2and
a temperature kT of 1.012.23
0.61 keV. The luminosity
in the 0.3 − 10.0 keV band was 5.5 × 1033
erg s−1,
which corresponds to an unabsorbed luminosity of 1.5 × 1034 erg s−1. The absorption column is
comparable to the value expected for NGC 346 (see paper II), but the temperature is one of the lowest we found, except for the extended source around HD 5980.
In the region of X-ray emission in NGC 346, 30 blue stars were detected by Massey, Parker & Garmany (1989): MPG 435, 342, 470, 368, 476, 396, 487, 417, 370, 471, 467, 495, 451, 330, 455, 454, 500, 445, 468, 375, 508, 395, 439, 371, 485, 561, 374, 366, 557, 400 (by increasing magnitude). Of these, 16 have known spectral types that we can use to convert magnitudes to bolometric lumi-nosities. The bolometric correction factors were taken from Humphreys & McElroy (1984). We can then compute the expected X-ray luminosi-ties using log(Lunabs
X ) = 1.13 log(LBOL) − 11.89
(Bergh¨ofer et al. 1997). If the X-rays were com-ing only from these 16 stars, we should expect a
total unabsorbed luminosity of 2 × 1033 erg s−1.
This represents only 13% of the detected luminos-ity. Several hypotheses could explain this discrep-ancy: a metallicity effect (Bergh¨ofer’s relation was determined for Galactic stars but SMC stars have weaker winds); X-ray emission from the other (un-classified) stars in this region, especially from the low-mass stars (see e.g. Getman et al. 2002); ad-ditional emission linked to interactions in binary systems (such as colliding winds); or an extended region of hot gas as the winds from the individual stars combine to form a cluster wind (Ozernoy, Genzel, & Usov 1997; Cant´o, Raga, & Rodr´ıguez 2000).
Chandrahas also observed X-ray emission from other very young (age < 5 Myr) stellar clusters, comparable to NGC 346. The Arches cluster in the Galactic Center region has been observed to have emission from several different regions, with a to-tal X-ray luminosity of Lunabs
X ∼5 × 10
35erg s−1
(Yusef-Zadeh et al. 2002). NGC 3603 has also been observed with Chandra (Moffat et al. 2002) and the total unabsorbed cluster luminosity is Lunabs
X ∼1 × 10
35 erg s−1 (though in this case it
is clear that a substantial fraction of the emission is from point sources in the cluster). For both the Arches and NGC 3603 clusters the fitted tem-perature of the X-ray emission is higher than the kT ∼ 1 keV value for NGC 346.
It is important to note that the calculations presented by Cant´o, Raga, & Rodr´ıguez (2000) and Raga et al. (2001) for the Arches cluster as-sume that there are about 60 stellar sources hav-ing mass-loss rates of 10−4 M
⊙ yr−1. If we take
the list of the bluest stars in NGC 346 given by Massey, Parker & Garmany (1989), the first 60 sources have spectral types in the range O3-B0. Assuming mass-loss rates for 8 SMC stars given by Puls et al. (1996) to be typical, then all stars in NGC 346 (excluding HD 5980) have mass-loss rates smaller than 10−4 M
⊙ yr−1. Only MPG
435 (O4III(n)(f)) has a mass loss rate as high as 9×10−5M
⊙ yr−1. However, the majority of these
60 stars are late O-type Main Sequence stars or cooler, which must have much smaller mass-loss rates (10−6 M
⊙ yr−1 or less), and they are more
widely distributed than in Arches. Thus, it is not
surprising that the X-ray luminosity is lower than in Arches, and, in fact, it may be too large to be explained solely with the cluster wind scenario. 4. The LBV HD 5980
Among all stars in NGC 346, HD 5980 is unique in its spectral variations in the last two decades, and warrants further examination. This eclips-ing binary (or triple system?) was classified as WN+OB before 1980, but its spectral type changed to WN3+WN4 in 1980–1983, and to WN6, with no trace of the companion in 1992. In 1994, this star underwent a LBV-type erup-tion, and presented at the same time a WN11 type. The eruption has now settled down and the spectrum is returning to its pre-eruption state (for more details see Koenigsberger et al. 2000, and references therein). We will label the 1994 eruptor as ‘star A’ and its companion as ‘star B’.
Chandra is the first X-ray telescope to detect this peculiar system, since the low spatial resolu-tion of previous X-ray observatories did not allow the distinction of HD 5980 from the extended emis-sion surrounding it. However, even with a 100 ks exposure, the data possess a rather low signal-to-noise ratio. The Chandra ACIS-I count rate of HD 5980 is only 2.98 ± 0.19 × 10−3cnts s−1 in the
0.3 − 10.0 keV band or about 300 counts in the total observation. This limits our ability to sensi-tively determine the spectral shape of the emission and to look for source variability. On the other hand, the relative faintness of the source means that photon pileup will not affect our analysis to any great degree.
We have extracted a spectrum of HD 5980 using the CIAO tool psextract. An annular background region around the source (but sufficiently far from it) was chosen, in order to eliminate contamina-tion from the surrounding extended emission (see § 5). This spectrum is shown in Fig. 5a. The best fit model to the spectrum has the follow-ing parameters: N (H) = 0.220.35
0.14 ×10 22
cm−2
and kT = 7.0413.35
4.35 keV for an absorbed mekal
model and N (H) = 0.280.44 0.19 ×10
22
cm−2 and
Γ=1.741.89
1.53 for an absorbed power-law. These
ab-sorption columns are consistent with the value expected for NGC 346 (see paper II) but part of
these columns are probably due to wind absorp-tion and/or absorpabsorp-tion from the 1994 ejecta. The derived observed X-ray luminosity of HD 5980 is LX= 1.3 × 1034erg s−1in the 0.3 − 10.0 keV band
for both models. Using the normalisation factor of the mekal model, we found a volume emission measure of ∼1057 cm−3 for this source.
We can compare HD 5980 with other WR stars detected in X-rays. The unabsorbed X-ray lu-minosity of HD 5980 is 1.7 × 1034 erg s−1 in the
0.3 − 10.0 keV energy range and 9 × 1033
erg s−1
in the ROSAT range, i.e. 0.2 − 2.4 keV. This places HD 5980 amongst the X-ray brightest single WN stars (Wessolowski 1996) and the brightest WR+OB binaries (Pollock 2002) of the Galaxy.
Two factors could explain this high X-ray lu-minosity. First, the fast wind from the post-eruptive phases (from 1300 km s−1 in 1994 to
∼2000 km s−1 in 2000) should now collide with
the slow wind (ejected with a velocity as low as 200 km s−1during the eruption, see e.g.
Koenigs-berger et al. 2000), which will increase the post-eruptive X-ray luminosity. X-ray emission from colliding ejecta has already been observed for an-other LBV, i.e. η Carinae. An elliptically shaped extended emission of 40′′×70′′(i.e. 0.4 pc×0.8 pc)
in size surrounds the star in the X-ray domain (Seward et al. 2001). It is correlated to the high velocity ejecta of the Homunculus nebula (Weis, Duschl, & Bomans 2001), which was created af-ter the last great eruption. For HD 5980, however, the ejecta from the 1994 eruption has not had sufficient time to form a detectable LBV nebula around the star, and no ‘Homunculus-like’ nebula from a previous eruption is visible around HD 5980 (see Fig. 2): any X-ray emission from the colliding ejecta should thus still be blended with the stellar emission from HD 5980 and that may contribute to explain the high X-ray luminosity. Unfortunately, no X-ray detection of HD 5980 before the 1994 eruption is available. A ROSAT all-sky survey im-age taken in 1990 (exposure number 933001) pro-vides only an upper limit of LX ∼1036 erg s−1 at
the position of HD 5980. Wang & Wu (1992) gave a 3σ upper limit of 2×1034
erg s−1 for all SMC
WR stars, except AB7. These limits are compat-ible with the detected luminosity of HD 5980 and do not allow us to quantitatively compare the
and post-eruption luminosities to confirm the pre-dicted luminosity enhancement.
The fact that HD 5980 is a close binary sys-tem containing two very massive stars, suggests that colliding winds may provide another source for the observed X-rays. We can thus estimate the expected X-ray luminosity. For the sys-tem, we assume that the stellar winds of the stars have returned to their pre-eruption values and we assume that ˙MA=1.4× 10−5 M⊙ yr−1,
v∞(A)=2500 km s−1 for the O-star, and ˙MB=
2×10−5 M
⊙ yr−1, v∞(B)=1700 km s−1 for the
WN Wolf-Rayet star (Moffat et al. 1998). Thus the momenta of each star’s wind are very similar, and the total wind kinetic energy is 5 × 1037
erg s−1. These wind parameters lead to a
momen-tum ratio of η = (MBv∞(B))/(MAv∞(A)) = 0.97
(Usov 1992) and the wind-wind collision shock will lie halfway between the stars. In this config-uration we expect about 1/6 of the wind kinetic energy to pass perpendicularly through the two shocks and be thermalised (Stevens, Blondin, & Pollock 1992), and because the systems are rela-tively close much of this energy will be radiated12.
Consequently, we would expect the X-ray lumi-nosity of HD 5980 to be ∼ 1036 erg s−1 or more,
rather than the observed value of ∼ 1034 erg s−1.
The discrepancy between the predicted and ob-served X-ray luminosities could be due to a range of factors, e.g. the winds may not be at the pre-eruption levels - lower mass-loss rates or lower wind velocities will translate to a lower X-ray lu-minosity. Radiative braking, wind clumping, or the eclipse of the colliding wind region by the thick wind of star B could also reduce the lumi-nosity. Alternatively, the wind momenta may not be so nearly equal, as a low value of η also tends to lower the luminosity.
Using the ephemeris of Sterken & Breysacher (1997), we have computed a phase of φ = 0.24 − 0.30 for our Chandra observation: this is close to the eclipse of star A by star B (φecl=0.36). If the
eccentricity is e ∼ 0.3, a simple adiabatic model
12In terms of the cooling parameter χ defined by Stevens, Blondin, & Pollock (1992), for the WN wind χ ≪ 1 and for the O-star wind χ ∼ 1.
predicts that the X-ray luminosity will vary in-versely with separation. We may expect a change in the intrinsic X-ray luminosity of a factor of ∼ 2 through the orbit and by ∼ 10% during the Chan-dra observation. We have detected a variation of apparently larger amplitude (factor ∼2) in the count rate of HD 5980 (see Fig. 5b). This still sug-gests we are seeing orbital variability associated with colliding winds, since the additional effects of changing absorption13
on the observed luminos-ity can either reduce or enhance this variabilluminos-ity. This observed variability is much more likely to be due to colliding winds than wind-blown bubble type emission, where no short timescale variability would be expected. The high fitted temperature of HD 5980 kT ∼ 7 keV also suggests that we are likely seeing colliding wind emission, as this tem-perature corresponds to a shock velocity of ∼ 2500 km s−1, comparable to the wind speed of star A
(v∞(A) above).
The detailed characteristics of the X-ray prop-erties of HD 5980 need to be studied further, in conjunction with detailed colliding wind models (c.f. the case of η Carinae; Pittard & Corcoran 2002). An XMM-Newton observation is scheduled at phases φ = 0.09 − 0.10, just after periastron: the comparison between these two datasets may enable us to detect further the signature of the col-liding wind region, and constrain it more precisely. Finally, we note that the unabsorbed X-ray luminosity of the central source in η Carinae, if we assume a distance of 2.3 kpc, varies be-tween (0.6 and 2.5)×1035
erg s−1 (Ishibashi et al.
1999). Such luminosities are much higher than for HD 5980.
5. The Extended Emission around HD 5980 Around the position of HD 5980 lies a region of bright extended X-ray emission. It was first detected by the Einstein Observatory (source IKT 18 in Inoue, Koyama, & Tanaka 1983), and sub-sequently observed by ROSAT (source [HFP2000]
13
The colliding wind region is viewed alternatively through the atmosphere of stars A and B. The absorption thus changes with the orbital phase.
14814 in Haberl et al. 2000). The Chandra image
reveals the extended emission in much more detail than previous X-ray observatories. The overall shape of the emission region is more or less rect-angular, with an extension to the northeast. It contains a few bright or dark arcs, but apart from these, the brightness is rather uniform, with no ob-vious limb-brightening. It has a size of 130′′×100′′,
i.e. 37pc×29pc at the SMC’s distance. HD 5980 lies towards the top center of the emission region. An X-ray bright filament extends from 9′′ west to
23′′south of HD 5980. An X-ray dark feature
ap-pears some 30′′at the southeast of HD 5980. The
total count rate of this source in the 0.3−10.0 keV energy range is 7.25 ± 0.17 × 10−2 cnts s−1. It is
the softest source present in the field.
The spectrum of this extended X-ray emis-sion was extracted in a rectangular aperture of ∼160′′×110′′ with a carefully chosen background
region, located as close as possible to the source and in a region on the ACIS-I1 CCD where there are no point sources. A circular region of di-ameter ∼5′′ containing HD 5980 was removed
from the extraction. As the X-ray emission is extended the reponse matrices were calculated us-ing the calcrmf and calcarf tools. The best fit (χ2
ν=1.03, Ndof=244) to this spectrum was an
absorbed mekal model with the following proper-ties (see Fig. 6): N (H) = 0.120.15
0.10×10 22 cm−2, kT = 0.660.68 0.64 keV and Z = 0.17 0.21 0.15Z⊙. The
observed flux in the 0.3 − 10.0 keV band was 3.35 × 10−13 erg cm −2 s−1, i.e. an observed
lu-minosity of 1.40 × 1035 erg s−1. The low value
of the absorption column, compared to HD 5980 and the cluster, suggests that this X-ray source lies between NGC 346 and the observer, but still in the SMC (see paper II).
Using the normalisation factor of the mekal model, we found a volume emission measure R
nenHdV of ∼3×1058 cm−3. Considering a
con-stant density for the hot gas, a sperical geometry of diameter ∼33 pc for the extended emission and assuming a pure H composition, the density of the hot gas is roughly 0.2 cm−3 and the total mass of
the hot gas is ∼100 M⊙.
14
This ROSAT source may encompass some of the X-ray emission from the cluster.
Using the available data, a deeper analysis was conducted, searching for the existence of a temper-ature gradient throughout the source area, with color and temperature maps. For color maps, we generated a set of narrow-band images, e.g. 0.3 − 0.9 keV, 0.9 − 1.2 keV and 1.2 − 2.0 keV, that enable us to distinguish the harder compo-nents from the softer parts. In addition, temper-ature maps were constructed from spectral fits in small regions of the SNR. The only obvious trend in both the color and temperature maps is the softness of the northeastern extension of the SNR. However, it is not possible to draw any firm conclusions regarding the presence of any temper-ature gradient.
We have also constructed images of the ex-tended emission in narrow energy bands, each correspond to a specific ion. The energy bands used for each ion are shown in Fig. 7. The data in each band were binned to obtain 4.9′′ ×4.9′′
pixels (see Fig. 7). In contrast to N132D (Behar et al. 2001), no significative differences between the morphology of highly ionized species (Mg10+,
Ne9+, Si12+, Fe19+) and lower ionization species
were detected.
5.1. The Nature of the Extended Emission Since its discovery the extended X-ray emis-sion has been attributed to a supernova remnant (SNR). Further evidence was subsequently ob-tained to support this hypothesis; for example, Ye, Turtle, & Kennicutt (1991) found a non-thermal radio source at this position that they called SNR 0057 − 7226. Using the lowest con-tours of Ye, Turtle, & Kennicutt (1991), the radio source is 56 pc×52 pc, slightly larger than the X-ray source.
Moreover, evidence of high velocity motions in this region were observed in the visible and UV ranges. Using echelle spectra centered on Hα, Chu & Kennicutt (1988) found clumps moving at vLSR ∼ 300 km s−1, redshifted from the main
quiescent component by +170 km s−1: only the
receding side of the expanding object is seen, sug-gesting a low density where the approaching side
arises. UV analyses with IUE (de Boer & Savage 1980), HST STIS (Koenigsberger et al. 2001) and FUSE (Hoopes et al. 2001) have also confirmed the presence of an expanding structure. While de Boer & Savage (1980) and Hoopes et al. (2001) found only absorptions at vLSR=+300 km s−1,
Koenigsberger et al. (2001) detected both ex-panding sides of the object, with components at vLSR = +21, +52, +300, +331 km s−1. However,
the absorptions at +21 and +52 km s−1 were
small, and need a future confirmation. The pres-ence of components from both approaching and receding sides of the expanding structure has now been confirmed by Danforth et al. (2002) from deep echelle spectroscopic integrations at optical wavelengths of the N66 region. Absorptions at any of these velocities are not seen in the spectra of Sk 80, a close neighbor of HD 5980 situated on the edge of the X-ray source. A fast expanding structure close to the position of HD 5980 is thus present.
Unfortunately, even the high-resolution HST WFPC2 images of the close neighborhood of HD 5980 do not show any clear nebula associated with the X-ray extended emission (see Fig. 2). A lot of filamentary structures - generally indicative of a SNR (Chen et al. 2000) - are seen throughout the field, but are not limited to the exact location of the X-ray source. There is thus no obvious Hα feature directly correlated with the extended X-ray emission.
Considering all the evidence (non-thermal emis-sion, high velocity expanding shell etc), the X-ray / radio source should thus be regarded as a SNR located in front of HD 5980 but still belonging to the SMC. The SNR hypothesis is further sup-ported by the comparison of the size of this fea-ture to its X-ray and radio fluxes (Mathewson et al. 1983).
As HD 5980 is projected more or less at the cen-ter of the diffuse X-ray emission and it underwent a LBV-type eruption in 1994 (which is unlikely to be the only one undergone by the system), it is tempting to associate the diffuse X-ray emission with HD 5980. To investigate whether this asso-ciation is likely, we first compare HD 5980 with η Carinae, the most well-known LBV that have
gone through multiple eruptions. η Car is also surrounded by bright, extended X-ray emission that appears to be assciated with the Carina Neb-ula: it is comparable in X-ray temperature (kT ), size and morphology to the X-ray emission around HD 5980 (Seward & Mitchell 1981; Fig.8). But it is generally assumed that this Nebula has been formed through the collective action of all stars of the Carina cluster, not only by the single action of η Carinae.
We have also searched for clues of an interac-tion between HD 5980 and its surroundings (e.g. a LBV nebula). Walborn (1978) noted evidence of such interaction: an arc of radius ∼ 30′′ centered
on the star is actually visible in Hα and [O iii] (but not in [S ii]). However, this arc is rather diffuse (see Fig. 2), indicating a lack of shock compres-sion. Moreover, its velocity coincides with that of the quiescent region, and no line-splitting re-gion is seen beginning at the arc and extending towards the star (Danforth et al. 2002): this arc is certainly not the rim of an expanding shell. This optical arc has different size, shape, and lo-cation from an X-ray bright arc (Fig. 2), therefore we conclude that they are probably two distinct features unrelated to each other. Echelle spectra of this area also do not show any enhancement of the [N ii]/Hα ratio which usually indicates the presence of circumstellar material surrounding the star. The comparison between X-rays and visible data thus do not show any clear indication of an interaction between HD 5980 and its environment. We also note that no circumstellar nebula pro-duced by HD 5980 could explain the redshifted UV absorptions in the spectra of the star, or the non-thermal radio emission from the extended source. The last possibility is that the source is indeed a SNR, but that HD 5980 is still directly respon-sible for it. Several authors have proposed the existence of a third component in HD 5980, ‘star C’, to interpret the visual light curve (Breysacher & Perrier 1991) and the presence of stable pho-tospheric absorptions in UV (Koenigsberger et al. 2000). To explain the existence of a SNR, an ad-ditional, fourth component of the system, ‘star D’, is needed. It would now be a compact object, a remnant of the star that exploded in SN long ago. If star C is much farther away than the system
A+B+D, then the high velocity redshifted UV absorptions could be produced by star C shining through the receding back wall of the SNR. The presence of a compact object might explain the ∼6 h variations seen in the lightcurve of HD 5980 by Sterken & Breysacher (1997). Accretion onto this compact object might also constitute another source of X-rays, thus contributing to understand the high X-ray emission from the system.
6. Summary
In this series of articles, we report the analysis of the Chandra data of N66, the largest star for-mation region of the SMC. In this first paper, we have focused on the most important objects of the field: the NGC 346 cluster and HD 5980.
The cluster itself is relatively faint, with a total luminosity of Lunabs
X ∼ 1.5 × 10
34 erg s−1 in the
0.3-10.0 keV energy range. Most of this emission seems correlated with the location of the brightest stars of the core of the cluster, but the level of X-ray emission probably cannot be explained solely by the emission from individual stars.
In this field lies another object of interest: HD 5980, a remarkable star that underwent a LBV-type eruption in 1994. Chandra is in fact the first X-ray telescope to detect HD 5980 indi-vidually. In X-rays, the star appears very bright, comparable only to the brightest WR stars in the Galaxy, but still fainter than η Carinae. The com-parison of our results with future X-ray observa-tions will enable us to better understand HD 5980: for example, phase-locked variations will be anal-ysed in the perspective of the colliding wind be-haviour of this binary, while other variations may be related to the recent LBV eruption. Follow-up observations are thus needed to complete the study of this system.
A bright, extended X-ray emission is seen to surround HD 5980. It is most probably due to a SNR whose progenitor is unknown. The spatial coincidence of this extended X-ray emission with the peculiar massive star suggest an association between these two objects. We also note the close resemblance of this X-ray emission to the Carina
Nebula in which the LBV η Carinae lies.
We are grateful to Dr Martin A. Guerrero Ron-cel for useful suggestions on data analysis tech-niques. Support for this work was provided by the National Aeronautics and Space Administra-tion through Chandra Award Number GO1-2013Z issued by the Chandra X-Ray Observatory Cen-ter, which is operated by the Smithsonian Astro-physical Observatory for and on behalf of NASA under contract NAS8-39073. Y.N. acknowledges support from the PRODEX XMM-OM and Inte-gral Projects, contracts P4/05 and P5/36 ‘Pˆole d’attraction Interuniversitaire (SSTC-Belgium) and from PPARC for an extended visit to the University of Birmingham. IRS and JMH also ac-knowledge support from PPARC. AFJM thanks NSERC (Canada) and FCAR (Quebec) for fi-nancial aid. This research has made use of the SIMBAD databse, operated at CDS, Strasbourg, France and NASA’s Astrophysics Data System Abstract Service.
REFERENCES
Arnaud, K. 1996, ASP Conf. Ser. 101, eds G. Ja-coby & J. Barnes, p.17
Behar, E., Rasmussen, A.P., Griffiths, R.G., Den-nerl, K., Audard, M., Aschenbach, B., & Brinkman, A.C. 2001, A&A, 365, L242
Bergh¨ofer, T.W., Schmitt, J.H.M.M., Danner, R., & Cassinelli, J.P. 1997, A&A, 322, 167
Breysacher, J. & Perrier, C. 1991, IAUS 143, 229 Cant´o, J., Raga, A.C., & Rodr´ıguez, L.F. 2000,
ApJ, 536, 896
Chen, C.-H. R., Chu, Y.-H., Gruendl, R.A., & Points, S.D. 2000, AJ, 119, 1317
Chu, Y.-H., & Kennicutt, R.C., Jr 1988, AJ, 95, 1111
Danforth et al., 2002, in preparation
de Boer, K.S., & Savage, B.D. 1980, ApJ, 238, 86 Getman, K.V., Feigelson, E.D., Townsley, L., Bally, J., Lada, C.J., & Reipurth B. 2002, to be published in ApJ, 575
Haberl, F., Filipovic, M. D., Pietsch, W., & Ka-habka, P. 2000, A&AS, 142, 41 (HFP2000) Haberl, F., & Sasaki, M. 2000, A&A 359, 573 Henize, K.G. 1956, ApJS, 2, 315
Hoopes, C.G., Sembach, K.R., Howk, J.C., & Blair, W.P. 2001, ApJ, 558, L35
Humphreys, R.M. & McElroy, D.B. 1984, ApJ, 284, 565
Inoue, H., Koyama, K., & Tanaka, Y. 1983, IAUS 101, 535 (IKT)
Ishibashi, K., Corcoran, M.F., Davidson, K., Swank, J.H., Petre, R., Drake, S.A., Damineli, A., & White, S. 1999, ApJ, 524, 983
Kaastra, J.S. 1992, An X-Ray Spectral Code for Optically Thin Plasmas (Internal SRON-Leiden Report, updated version 2.0)
Kahabka, P., Pietsch, W., Filipovic, M. D., & Haberl, F. 1999, A&AS, 136, 81
Koenigsberger, G., Georgiev, L., Barb`a, R., Tzve-tanov, Z., Walborn, N.R., Niemela, V.S., Mor-rell, N., & Schulte-Ladbeck, R. 2000, ApJ, 542, 428
Koenigsberger, G., Georgiev, L., Peimbert, M., Walborn, N.R., Barb`a, R., Niemela, V.S., Mor-rell, N., Tsvetanov, Z., & Schulte-Ladbeck, R. 2001, AJ, 121, 267
Massey, P., Parker, J.W., & Garmany, C.D. 1989, AJ, 98, 1305
Mathewson, D.S., Ford, V.L., Dopita, M.,A., Tuohy, I.R., Long, K.S., & Helfand, D.J. 1983, ApJS, 51, 345
Mathewson, D.S., Ford, V.L., & Visvanathan, N. 1986, ApJ 301, 664
Moffat, A.F.J., Marchenko, S.V., Bartzakos, P., Niemela, V.S., Cerruti, M.A., Magalhaes, A.M., Balona, L., St-Louis, N., Seggewiss, W., & Lamontagne, R. 1998, ApJ, 497, 896 Moffat, A.F.J., et al., 2002, ApJ, (in press) Naz´e et al., in preparation (paper II)
Ozernoy, L.M., Genzel, R., & Usov, V.V. 1997, MNRAS, 288, 237
Pittard, J.M., & Corcoran, M.F. 2002, A&A, 383, 636
Pollock, A.M.T. 2002, Interacting winds from massive stars, eds A.F.J. Moffat & N. St Louis, ASP Conf. Series, 260, p 363
Puls, J., Kudritzki, R.-P., Herrero, A., Pauldrach, A.W.A., Haser, S.M., Lennon, D.J., Gabler, R., Voels, S.A., Vilchez, J.M., Wachter, S., & Feld-meier, A. 1996, A&A, 305, 171
Raga, A.C., Vel´azquez, P.F., Cant´o, J., Masciadri, E., & Rodr´ıguez, L.F. 2001, ApJ, 559, L33 Sasaki, M., Haberl, F., & Pietsch, W. 2000,
A&AS, 147, 75
Schmidtke, P.C., Cowley, A.P., Crane, J.D., Tay-lor, V.A., McGrath, T.K., Hutchings, J.B., & Crampton, D. 1999, AJ, 117, 927
Seward, F.D., Butt, Y.M., Karovska, M., Prest-wich, A., Schlegel, E.M., & Corcoran, M. 2001, ApJ, 553, 832
Sterken, C. & Breysacher, J. 1997, A&A, 328, 269 Stevens I.R., Blondin J.M., Pollock, A.M.T. 1992,
ApJ, 386, 265
Townsley, L.K., Broos, P.S., Garmire, G.P., & Nousek, J.A. 2000, ApJ, 534, L139
Usov, V.V. 1992, ApJ, 389, 635 Walborn, N.R. 1978, ApJ, 224, L133
Walborn, N.R., Howarth, I.D., Lennon, D.J., Massey, P., Oey, M.S., Moffat, A.F.J., Skalkowski, G., Morrell, N.I., Drissen, L., & Parker, J.W. 2002, AJ, 123, 2754
Wang, Q. & Wu, X. 1992 ApJS, 78, 391
Weis, K., Duschl, W.J., & Bomans, D.J. 2001, A&A, 367, 566
Wessolowski, U. 1996, MPE report No. 263, 75 Ye, T., Turtle, A.J., & Kennicutt, R.C., Jr. 1991,
MNRAS, 249, 722
Yokogawa, J., Imanishi, K., Tsujimoto, M., Nishi-uchi, M., Koyama, K., Nagase, F., & Corbet, R.H.D. 2000, ApJS, 128, 491
Yusef-Zadeh, F., Law, C., Wardle, W., Wang, Q.D., Fruscione, A., Lang, C.C., Cotera A. 2002, ApJ, 570, 665
This 2-column preprint was prepared with the AAS LATEX macros v5.0.
Fig. 1.— Chandra color image of the region close to HD 5980. Three energy bands were used to create this color image : red corresponds to 0.3-1.0 keV, green to 0.3-1.0-2.0 keV and blue to 2.0-10.0 keV. Before combination, the images were smoothed by convolution with a gaussian of σ=1′′.
Fig. 2.— HST WFPC2 Hα data of the region sur-rounding HD 5980 (which is the star located at the center of the PC chip). The X-ray data overlaid as contours on the HST image were first binned by a factor of 2 and then smoothed by convolution with a gaussian of σ=2′′.
Fig. 3.— The Chandra X-ray data of the NGC 346/HD 5980 region superimposed on the DSS image, showing the extended nature of the X-ray source associated with the cluster. The lo-cation of the brightest stars in the cluster are in-dicated. The X-ray data were first binned by a factor of 2 and then smoothed by convolution with a gaussian of σ = 2′′)
Fig. 4.— The X-ray spectrum of the NGC 346 cluster, shown along with the best-fit absorbed mekal model, with N (H) = 0.42 × 1022cm−2 and
kT = 1.01 keV (see § 3).
Fig. 5.— a. The Chandra X-ray spectrum of HD 5980, shown along with the best-fit absorbed mekal model, with N (H) = 0.22 × 1022
cm−2 and
kT = 7.04 keV. b. The Chandra X-ray lightcurve of HD 5980, in the 0.3-10 keV range and with 10 bins of 10 ks each.
Fig. 6.— The Chandra X-ray spectrum of the extended emission around HD 5980, shown along with the best-fit absorbed mekal model, with N (H) = 0.12 × 1022 cm−2, kT = 0.66 keV and
Z = 0.17Z⊙(see § 5).
Fig. 7.— Narrow-band X-ray images of the ex-tended emission surrounding HD 5980. Each im-age is labeled with the principal line-emitting ion and the energy range used to generate the image.
Fig. 8.— X-ray images of the Carina Nebula (24 ks ; ROSAT PSPC observation #900176) com-pared to the extended emission around HD 5980. If we assume a distance of 2.3 kpc for the Ca-rina Nebula and 59 kpc for the SMC, both images are ∼55 pc×55 pc. ROSAT PSPC support rings and Chandra ACIS-I CCD gaps can be spotted on these images. Both images were smoothed by convolution with a gaussian (σ=15′′ for the
Ca-rina Nebula and σ=2′′after binning by a factor of
